INP is the ecommerce interaction budget
Checkout can feel broken after LCP passes. Field INP shows where tags, hydration and handlers delay the next paint — and what leaders should measure.
Insights
Thinking in public
Notes on ecommerce engineering, frontend craft, engineering leadership, practical AI and the changing search landscape.
-
-
We shipped more code. We did not ship less risk.
AI-assisted delivery can raise merge volume while rework and near-term churn remain hidden. What engineering leaders should measure instead.
-
Peak season is an engineering deadline, not a marketing one
Black Friday readiness starts before campaign launch: load-test checkout journeys, control deploys and govern script cost before peak traffic arrives.
-
Agentic commerce and the storefront engineering test
Agentic commerce fails on storefront basics before protocols matter — reach, readable facts, aligned catalog truth, and only then UCP, ACP or MCP integration.
-
The review bar after AI
How pull requests stay reviewable when AI-generated diffs look finished — accountability questions, the cost of rubber-stamping, and how leaders keep the merge bar honest.
-
Third-party scripts are an engineering leadership problem
Why third-party scripts become a performance, reliability and ownership problem — and how ecommerce leaders can govern their cumulative cost.
-
Practical AI in engineering — judgment over volume
Why engineering teams get more from AI when they treat it as leverage for judgment — not a shortcut past craft, review and accountability.
-
AEO and GEO for Ecommerce: What Has Actually Changed?
What AEO, GEO and agentic commerce change for ecommerce teams — and what remains crawlability, entities and trustworthy structured data.
-
Staying close to the craft
Why engineering leadership is stronger when it stays grounded in how software is designed, built and maintained — without becoming a delivery bottleneck.
